{"schema_version":"newruntime-agent-readable-v0.2","type":"post","stable_id":"post:anthropic-claude-cryptographic-weaknesses","slug":"anthropic-claude-cryptographic-weaknesses","title":"Claude Mythos Moves Cryptanalysis Into the Verification Bottleneck","description":"Anthropic's cryptography research shows a frontier model finding HAWK and reduced-round AES attacks quickly, while human validation and disclosure become the scarce production step.","retrieval_nugget":"Anthropic's cryptography research shows a frontier model finding HAWK and reduced-round AES attacks quickly, while human validation and disclosure become the scarce production step. Anthropic's cryptography post is not a normal \"AI found bugs\" story. The public evidence spans 5 source links and points to a narrower bottleneck: a frontier model can generate cryptographic attack ideas quickly, but human validation.","status":"published","published_at":"2026-07-30","updated_at":"2026-07-30","record_date":"2026-07-30","date_kind":"published_at","topics":["security","evals","frontier-models","verification","ai-research"],"source_urls":["https://x.com/AnthropicAI/status/2082153297670992134","https://x.com/AnthropicAI/status/2082153309553463600","https://www.anthropic.com/research/discovering-cryptographic-weaknesses","https://www-cdn.anthropic.com/e8d50c167ad47beeb03d6109a4a484be95cb38ea/hawk_key_recovery.pdf","https://www-cdn.anthropic.com/c88771e1bf5ee8885349eed05e5484c0e5f7e02b/aes_mobius_bridge.pdf"],"visuals":[{"id":"anthropic-claude-cryptographic-weaknesses-nano-banana","kind":"editorial-diagram","role":"hero","src":"https://newruntime.com/images/posts/anthropic-claude-cryptographic-weaknesses-nano-banana.webp","alt":"Hand-drawn whiteboard diagram showing abstract cipher puzzles moving through AI search workers, human proof review, safe disclosure, and a blocked unsafe path.","caption":"Anthropic's result is less a deployment break than a warning that AI-generated research can outrun human validation capacity.","credit":"New Runtime synthesis from public source inspection","source_url":"https://www.anthropic.com/research/discovering-cryptographic-weaknesses","generated_with":"nano-banana-style-imagegen","width":1600,"height":900,"legend":[{"label":"HAWK","description":"Anthropic says Claude Mythos Preview improved the best-known attack on the HAWK post-quantum signature candidate after about 60 hours of work."},{"label":"Reduced AES","description":"The post says Mythos found a faster attack on a seven-round AES variant, improving prior approaches by 200-800x depending on measurement."},{"label":"Validation","description":"Anthropic reports that human researchers spent much longer validating and writing up the AES result than the model spent finding the idea."},{"label":"Disclosure","description":"The work was shared after consultation and responsible disclosure, and the post says it does not affect production systems today."}]}],"telegram_message_id":2816,"telegram_url":"https://t.me/qwgai/2816","telegram_message_ids":[2815,2816],"telegram_delivery_mode":"media_then_text","telegram_media_url":"https://t.me/qwgai/2815","routes":{"html":"https://newruntime.com/posts/anthropic-claude-cryptographic-weaknesses/","markdown":"https://newruntime.com/posts/anthropic-claude-cryptographic-weaknesses.md","json":"https://newruntime.com/posts/anthropic-claude-cryptographic-weaknesses.json"},"source_format":"markdown","next_reads":[{"type":"topic","path":"/topics/evals/","reason":"Explore the evals topic hub.","url":"https://newruntime.com/topics/evals/","title":"Agent evals - New Runtime","media_type":"text/html"},{"type":"topic","path":"/topics/security/","reason":"Explore the security topic hub.","url":"https://newruntime.com/topics/security/","title":"Security - New Runtime","media_type":"text/html"},{"type":"related_material","path":"/posts/langchain-reviewbench-review-agent-evals/","reason":"Shares evals and verification.","url":"https://newruntime.com/posts/langchain-reviewbench-review-agent-evals/","title":"ReviewBench Turns Code Review Into An Agent Eval","media_type":"text/html"},{"type":"related_material","path":"/posts/deepsecbench-security-agent-economics/","reason":"Shares evals and security.","url":"https://newruntime.com/posts/deepsecbench-security-agent-economics/","title":"DeepsecBench Makes Security Agents a Cost/Recall Tradeoff","media_type":"text/html"},{"type":"related_material","path":"/posts/agentic-sdlc-software-factory-loop/","reason":"Shares evals.","url":"https://newruntime.com/posts/agentic-sdlc-software-factory-loop/","title":"A Software Factory Connects Agents Through Verified Outcomes","media_type":"text/html"}]}